You are on page 1of 44

April 2009/ John Wgter

Note 2.1 Fatigue design based on S-N data

General .............................................................................................................................................2 Nominal stress..................................................................................................................................2 Structural stress ................................................................................................................................6 Notch stress ......................................................................................................................................7 Hot spot stress ..................................................................................................................................8 Stress range ....................................................................................................................................10 Fatigue resistance according to DNV-RP-C203 ............................................................................12 Introduction................................................................................................................................12 General (2.4.1) ...........................................................................................................................12 Failure criterion inherent the S-N curves (2.4.2) .......................................................................12 S-N curves and joint classification (2.4.3).................................................................................13 S-N curves in air (2.4.4).............................................................................................................14 S-N curves in seawater with cathodic protection (2.4.5) ...........................................................15 S-N curves for tubular joints (2.4.6) ..........................................................................................16 S-N curves for free corrosion (2.4.9) .........................................................................................17 Effect of fabrication tolerances (2.6) .........................................................................................18 Stress Concentration Factors (3) ................................................................................................18 Stress concentration factors for tubular butt weld connections (3.3.7)......................................19 DNV Appendix A ......................................................................................................................23 References......................................................................................................................................44

Fatigue design based on S-N data


General
Fatigue analysis of welded components is based on the long term distribution of stresses from either a known load history or from design rules. The capacity with respect to the fatigue strength is characterized by S-N curves, which give the relationship between the stress ranges applied to a given detail and the number of constant amplitude load cycles to failure. Depending on the kind of stresses used in the calculation, the fatigue assessment may be categorised by the so-called nominal stress approach, hot spot stress approach and notch stress approach. The three stresses are defined as follows: Nominal stress A general stress in a structural component calculated by beam theory based on the applied loads and the sectional properties of the component. The sectional properties are determined at the section considered (i.e. the hot spot location) by taking into account the gross geometric changes of the detail (e.g. cut-outs, tapers, haunches, brackets, changes of scantlings, misalignments, etc.). The nominal stress can also be calculated using a coarse mesh FE analysis or an analytical approach. A local stress at the hot spot (point with highest stress) where cracks may be initiated. The hot-spot stress takes into account the influence of structural discontinuities due to the geometry of the connection, but excludes the effects of the weld itself. The hot spot stress is also referred to as structural stress or geometric stress. A peak stress at the root/toe of a weld or notch taking into account stress concentrations due to the effects of structural geometry as well as the presence of the weld.

Hot spot stress

Notch stress

For this course the emphasis is on the two first approaches, the nominal stress approach and the hot spot stress approach, as these approaches currently dominate practical design work. The notch stress approach is used in special cases where reliable results cannot be obtained based on the two first methods.

Nominal stress
Traditional fatigue analysis of welded components is based on the use of nominal stresses and catalogues of classified details. A particular type of detail is assigned to a particular fatigue class with a given S-N curve.

In general, nominal stresses are determined using the beam theory

nom =
where

F M + y A I

(1)

F is axial force A is area of cross section M is bending moment I is moment of inertia of the cross section y is distance from centroid to the point considered

An example of the determination of the nominal stress at a welded attachment is shown in Figure 1. Here the nominal stress is simply determined as the beam stress in the region containing the weld detail, but without considering any influence of the attachment on the stress distribution.

Figure 1

Nominal stress in beam-like component

In some cases it might also be necessary to include the effect of certain macro-geometric features as well as stress fields in the vicinity of concentrated forces and reaction forces when the nominal stress is determined. Examples of welded structures that contain macro-geometrical forms that are not included in the classified detail in design codes are given in Figure 2. In all cases the stress field is altered as compared to the stresses calculated using elementary stress analysis formulas.

Figure 2

Examples of macro-geometric effects

Examples of stress fields in the vicinity of concentrated loads and reaction forces are shown in Figure 3.

Figure 3

Modified (local) nominal stresses near concentrated loads

Another example shown in Figure 4 further illustrates the correct determination of the stress to be considered in the fatigue calculation. In b) the stress concentration from the hole is a macrogeometric effect that must be accounted for, giving the relevant local stress equal to SCFmacro nom where SCFmacro is the stress concentration factor due to the hole, i.e.

local =

nom

SCFmacro nom

(2)

This local stress shall be used together with the relevant S-N curves dependent on the joint classification.

Figure 4

Local stresses

Summarizing these findings we can state that The joint classification and the corresponding S-N curve used in association with nominal stresses take the local stress concentration created by the joint itself and by the weld profile into account. The design stress can typically be determined as the nominal stress adjacent to the weld under consideration. However, in cases where macro-geometric features or local loading create a stress pattern different to that determined by simple stress analysis formulas, the relevant stress for the fatigue calculation is the local stress determined in Equation (2).

Another aspect that may require special attention is the misalignment included in the fatigue strength from test specimens.
Unfortunately, in most data reproduced in relation to fatigue strength, the built-in misalignment is not quantified. In some design standards it is assumed that the fatigue strength of a considered joint detail includes normal fabrication tolerances for the actual type of detail, while in other design standards, it is required to include a specific stress concentration factor to take realistic fabrication tolerances into consideration. Generally, this aspect is most critical for plated structures for which most modern design standards today require the inclusion of an SCF, typically given by

SCF = 1 + 3

e t

(3)

where e is the eccentricity and t the plate thickness.

Structural stress
Structural stresses accounts for any stress created by the considered detail exclusive of the nonlinear stress field due to the notch at the weld toe. Stresses in shell structures which are determined based on the theory of shells are structural stresses. In the shell theory stresses are linearly distributed across the plate thickness and consists of two parts: membrane stress and shell bending stress, see Figure 5. It should be noted that membrane stresses and local bending stresses also occur in e.g. plated structures, and for ease of reference we still denote these stress components shell stresses.

Figure 5

Shell stresses

The structural stress includes all stress raising effects of the structural detail, but excludes the stress concentration due to the local weld profile itself. Figure 6 shows examples of structural details together with the structural stress distribution.

Figure 6

Structural details and structural stress

Notch stress
The notch stresses accounts for any stress created by the considered detail inclusive of the nonlinear stress field due to the notch at the weld toe. It is composed of the sum of the structural stress and the nonlinear peak stress, see Figure 7.

Figure 7

Local notch stress at a weld toe

The stress components of the notch stress ln are:

mem membrane stress ben shell bending stress nlp non-linear stress peak stress
If a refined stress analysis method that gives a non-linear stress distribution is used, the stress components can be separated by the following method: The membrane stress mem is equal to the average stress calculated through the thickness of the plate. It is constant through the thickness. The shell bending stress ben is linearly distributed through the thickness of the plate. It is found by drawing a straight line through the point O where the membrane stress intersects the mid-plane of the plate. The gradient of the shell bending stress is chosen such that the remaining non-linearly distributed component is in equilibrium. The non-linear stress peak nlp is the remaining component of the stress.

The stress components can be separated analytically for a given stress distribution ( x ) as follows

Figure 8

Separation of stress components

Hot spot stress


Fatigue design of more complicated details is normally based on structural hot spot stresses because the nominal stress approach has severe limitations. The nominal stress approach ignores the actual dimensional variations of a particular structural detail, and often the shape of the welded component is so complex that the determination of the nominal stress is difficult or impossible. Also, it requires a specific, experimentally determined S-N curve for the detail considered. Thus, for details with no available design S-N curves, the designer needs a more general design tool based on structural hot spot stresses. Hot spots are points with the highest stresses, and while structural stresses are the general term for the shell stresses in the structure, the hot spot stress is the value of the structural stress on the surface at the hot spot of the component which is to be assessed. The structural hot spot stress approach is generally applicable for the fatigue design of welded plate, shell and tubular structures where cracks at the weld toe are critical. If critical cracks grow from the weld root special care should be taken. Thus, in Ref. /3/ the hot spot method is therefore limited to the assessment of the weld toe, i.e. cases a to e in Figure 9, while it is not immediately applicable in cases where cracks grow from the weld root and propagate through the weld metal, i.e. cases f to i in Figure 9.

Figure 9

Various locations of crack propagation in welded joints

However, in Ref. /3/ it is also noted that:


The method of structural hot spot stress may be extended to the assessment of spots of the welded joint susceptible to fatigue cracking other than on plate surface, e.g. on a fillet weld root. In this case, structural hot spot stress on surface is used as an indication and estimation of the stress for the spot in consideration. The S-N curves or structural hot spot stress concentration factors used for verification in this case depend largely on geometric and dimensional parameters and are only valid within the range of these parameters.

The structural stress includes all stress raising effects of the structural detail, but excludes the stress concentration due to the local weld profile itself. Thus, when analyzing a detail the designer must be able to separate the appropriate stress components to derive the relevant stresses for the hot spot design. A detailed FE analysis of a welded joint may result in the stresses shown as total stresses in Figure 10.

Stress distribution along surface normal to weld

Notch stress Geometric stress Nominal stress

Stress Distributions through the thickness of the plate/tube wall


Notch Stress Zone Geometric stress zone Nominal Stress Zone

Section A-A

Section B-B

Section C-C

Figure 10

Definition of structural hot spot stress

The stresses far away from the weld are the nominal stresses. Closer to the weld the stress raising effect of the structural detail becomes important and in the close vicinity of the weld the stress concentration due to the weld itself generates very high notch stresses.

In a hot spot stress design the appropriate stresses are those that catch the stress concentration from the structural detail, but not the stress concentration from the weld itself. The structural hot spot stresses thus defined can be determined by using reference points and extrapolation to the weld toe at the hot spot in consideration as shown in Figure 10. The stress extrapolation procedure given above is the classical approach to derive the hot spot stresses, but other procedures based on a single stress value in a read out points close to the weld toe are also used (Method B in Ref. /4/). For a more complex state of stress, e.g. a biaxial stress state in a plate, the question arises in relation to which stress component should be considered. Traditionally, the principal stress has always been considered a significant parameter for the analysis of fatigue crack growth. At least as long as the principal stress is essentially normal to the weld. Some guidelines recommend using the normal to the weld stress component in cases where the principal stress tends to become parallel with the weld, while Ref. /4/ gives guidance to calculate an effective hot spot stress based on a more complex procedure.

Stress range
The stress range, see Figure 11, is the main parameter to be determined for fatigue analysis. In the case of constant amplitude loading, Figure 11 (a), the stress range is defined as
= max min

(4)

This definition of stress range applies for any of the three stress categories defined previously. In many real structures, in particular for welded structures, variable amplitude loading, Figure 11 (b) is more common than constant amplitude loading.

Figure 11

Constant (a) and variable (b) amplitude stress histories

Fatigue analysis is based on the cumulative effect of all stress range occurrences during the design life. A stress range occurrence table or a stress range spectrum must therefore be produced from the stress history by an appropriate counting method, typically Rainflow counting. The stress range spectrum, Figure 12, is a representation of stress ranges and the associated number of cycles. Using

10

the stress range spectrum in combination with Miners rule allows a direct calculation of the cumulative fatigue from the considered stress history.

Figure 12

Stress range spectrum

11

Fatigue resistance according to DNV-RP-C203


Introduction
In the following the design S-N curves and procedures as recommended in Ref. /4/ are used as an example of a modern state of the art design code for fatigue. The text in this section is a copy of the corresponding parts of C203 and has not been edited or changed in any way (although not all clauses of C203 have been reproduced here), and the original references to figures and tables (in Ref. /4/) have been maintained for easy cross reference.

General (2.4.1)
The fatigue design is based on use of S-N curves, which are obtained from fatigue tests. The design S-N curves which follows are based on the mean-minus-two-standard-deviation curves for relevant experimental data. The S-N curves are thus associated with a 97.6% probability of survival.

Failure criterion inherent the S-N curves (2.4.2)


Most of the S-N data are derived by fatigue testing of small specimens in test laboratories. For simple test specimens the testing is performed until the specimens have failed. In these specimens there is no possibility for redistribution of stresses during crack growth. This means that most of the fatigue life is associated with growth of a small crack that grows faster as the crack size increases until fracture. For details with the same calculated damage, the initiation period of a fatigue crack takes longer time for a notch in base material than at a weld toe or weld root. This also means that with a higher fatigue resistance of the base material as compared with welded details, the crack growth will be faster in base material when fatigue cracks are growing. For practical purpose one defines these failures as being crack growth through the thickness. When this failure criterion is transferred into a crack size in a real structure where some redistribution of stress is more likely, this means that this failure criterion corresponds to a crack size that is somewhat less than the plate thickness. The tests with tubular joints are normally of a larger size. These joints also show larger possibility for redistribution of stresses as a crack is growing. Thus a crack can grow through the thickness and also along a part of the joint before a fracture occur during the testing. The number of cycles at a crack size through the thickness is used when the S-N curves are derived. As these tests are not very different from that of the actual behaviour in a structure, this failure criterion for S-N curves for tubular corresponds approximately to the thickness at the hot spot (chord or brace as relevant).

12

S-N curves and joint classification (2.4.3)


For practical fatigue design, welded joints are divided into several classes, each with a corresponding design S-N curve. All tubular joints are assumed to be class T. Other types of joint, including tube to plate, may fall in one of the 14 classes specified in Table 2-1, Table 2-2 and Table 2-3, depending upon: the geometrical arrangement of the detail the direction of the fluctuating stress relative to the detail the method of fabrication and inspection of the detail. Each construction detail at which fatigue cracks may potentially develop should, where possible, be placed in its relevant joint class in accordance with criteria given in Appendix A. It should be noted that, in any welded joint, there are several locations at which fatigue cracks may develop, e. g. at the weld toe in each of the parts joined, at the weld ends, and in the weld itself. Each location should be classified separately.

The fatigue strength of welded joints is to some extent dependent on plate thickness. This effect is due to the local geometry of the weld toe in relation to thickness of the adjoining plates. See also effect of profiling on thickness effect in section 7.2. It is also dependent on the stress gradient over the thickness. Reference is made to Appendix D, Commentary. The thickness effect is accounted for by a modification on stress such that the design S-N curve for thickness larger than the reference thickness reads:

13

In general the thickness exponent is included in the design equation to account for a situation that the actual size of the structural component considered is different in geometry from that the S-N data are based on. The thickness exponent is considered to account for different size of plate through which a crack will most likely grow. To some extent it also accounts for size of weld and attachment. However, it does not account for weld length or length of component different from that tested such as e. g. design of mooring systems with a significant larger number of chain links in the actual mooring line than what the test data are based on. Then the size effect should be carefully considered using probabilistic theory to achieve a reliable design, see Appendix D, Commentary.

S-N curves in air (2.4.4)


S-N curves for air environment are given in Table 2-1 and Figure 2-6. The T curve is shown in Figure 2-8. In the low cycle region the maximum stress range is that of the B1 curve as shown in Figure 2-6. However, for offshore structures subjected to typical wave and wind loading the main contribution to fatigue damage is in the region N > 106 cycles and the bilinear S-N curves defined in Table 2-1 can be used.

14

S-N curves in seawater with cathodic protection (2.4.5)


S-N curves for seawater environment with cathodic protection are given in Table 2-2 and Figure 2-7. The T curve is shown in Figure 2-8. For shape of S-N curves see also comment in 2.4.4.

15

S-N curves for tubular joints (2.4.6)


S-N curves for tubular joints in air environment and in seawater with cathodic protection are given in Table 2-1, Table 2-2 and Table 2-3.

16

S-N curves for free corrosion (2.4.9)


S-N curves for free corrosion, i.e. without corrosion protection, are given in Table 2-3. See also Commentary section for consideration of corrosion protection of connections in the splash zone and inside tanks in FPSOs.

17

Effect of fabrication tolerances (2.6)


Normally larger fabrication tolerances are allowed in real structures than that accounted for in the test specimens used to derive S-N data, ref. DNV OS-C401; Fabrication and Testing of Offshore Structures. Therefore, additional stresses resulting from normal fabrication tolerances should be included in the fatigue design. Special attention should be given to the fabrication tolerances for simple butt welds in plates and tubulars as these give the most significant increase in additional stress. Stress concentration factors for butt welds are given in section 3.1.2 and at tubular circumferential welds in section 3.3.7.

Stress Concentration Factors (3)


Stress concentration factors for plated structures (3.1) General (3.1.1)
A stress concentration factor may be defined as the ratio of hot spot stress range over nominal stress range.

Stress concentration factors for butt welds (3.1.2)


The eccentricity between welded plates may be accounted for in the calculation of stress concentration factor. The following formula applies for a butt weld in an unstiffened plate or for a pipe butt weld with a large radius:

The stress concentration for the weld between plates with different thickness in a stiffened plate field may be derived from the following formula:

18

Stress concentration factors for tubular butt weld connections (3.3.7)


Due to less severe S-N curve for the outside weld toe than the inside weld root, it is strongly recommended that tubular butt weld connections subjected to axial loading are designed such that any thickness transitions are placed on the outside (see Figure 3-8). For this geometry, the SCF for the transition applies to the outside. On the inside it is then conservative to use SCF = 1.0. Thickness transitions are normally to be fabricated with slope 1:4. Stress concentrations at tubular butt weld connections are due to eccentricities resulting from different sources. These may be classified as concentricity (difference in tubular diameters), differences in thickness of joined tubulars, out of roundness and centre eccentricity, see Figure 3-10 and Figure 3-11. The resulting eccentricity may be conservatively evaluated by a direct summation of the contribution from the different sources. The eccentricity due to out of roundness normally gives the largest contribution to the resulting eccentricity . It is conservative to use the formula for plate eccentricities for calculation of SCF at tubular butt welds. The effect of the diameter in relation to thickness may be included by use of the following formula, provided that T/t 2:

19

This formula also takes into account the length over which the eccentricity is distributed: L, ref. Figure 3-9 and Figure 3-8. The stress concentration is reduced as L is increased and or D is reduced. It is noted that for small L and large D the last formula provides stress concentration factors that are close to but lower than that of the simpler formula for plates. The transition of the weld to base material on the outside of the tubular can normally be classified to S-N curve E. If welding is performed in a horizontal position it can be classified as D. This means that the pipe would have to be rotated during welding. Equation (3.3.4) applies for the outside tubular side shown in Figure 3.8. For the inside the following formula may be used:

If the transition in thickness is on the inside of the tubular and the weld is made from both sides, equation (3.3.4) may be applied for the inside weld toe and equation (3.3.5) for the outside weld toe. If the transition in thickness is on the inside of the tubular and the weld is made from the outside only, the following formulae may be used for the inside weld root:

And equation (3.3.5) may be applied for the outside weld toe.

20

In tubulars, the root side of welds made from one side is normally classified as F3. This requires good workmanship during construction, in order to ensure full penetration welds, and that work is checked by non-destructive examination. It may be difficult to document a full penetration weld in most cases due to limitations in the non-destructive examination technique to detect defects in the root area. The F3 curve can be considered to account for some lack of penetration, but it should be noted that a major part of the fatigue life is associated with the initial crack growth while the defects are small. This may be evaluated by fracture mechanics such as described in BS 7910 Guidance on Methods for Assessing the Acceptability of Flaws in Fusion Welded Structures, ref /7/. Therefore, if a fabrication method is used where lack of penetration is to be expected, the design S-N curves should be adjusted to account for this by use of fracture mechanics. For global moments over the tubular section it is the nominal stress derived at the outside that should be used together with an SCF from equation (3.3.4) for calculation of hot spot stress for fatigue assessment of the outside weld toe. The nominal stress on the inside should be used for assessment of fatigue cracks initiating from the inside.

21

22

DNV Appendix A

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

References
Ref. /1/ IACS Recommendation No. 56 Fatigue Assessment of Ship Structures. Recom. 56.1, July 1999. Niemi, E.: Stress Determination for Fatigue Analysis of Welded Components. IIS/IIW-1221-93. Abington Publishing, Cambridge, England 1995. Hobbacher, A.: Recommendation for Fatigue Design of Welded Joints and Components. IIW document XIII-1965-03/XV-1127-03. Update July 2004. DNV-RP-C203: Fatigue Design of Offshore Steel Structures. October 2008. Det Norske Veritas. Niemi, E.: Structural Stress Approach to Fatigue Analysis of Welded Components Designers Guide. IIS/IIW-XIII-1819-00. Final draft subject to editorial amendments. Last modified August 12, 2001.

Ref. /2/

Ref. /3/

Ref. /4/

Ref. /5/

44